Files
seismo-relay/docs
serversdownandClaude Opus 5 98cbdad489 docs(series4): Repeat is folded into Action -- and [6] is not the repeat flag
A capture that changed ONLY the Repeat Daily checkbox (pull schedule, disable
repeat, push schedule) moved exactly one byte in the entire session:

    record @0   Action  0x03 -> 0x02

Record 2 unchanged, [6] unchanged, and 0xDA / 0x68 / 0x82 / 0x71 / 0x94 / 0x8D all
byte-identical.

Two corrections to the previous commit:

  * [6] is NOT the repeat flag.  That was the field I predicted this capture would
    isolate; it stayed 2 and 16.  Still unidentified.
  * The label "Action" on [0] was too simple -- it carries repeat behaviour too.

Leading hypothesis: 2 and 3 are the two setup-bearing start actions the firmware
names, with repeat selecting between them --
    0 = DUTYCYCLE_CALLHOME
    2 = DUTYCYCLE_START_MONITOR_WITH_SETUP                 (repeat off)
    3 = DUTYCYCLE_START_MONITOR_WITH_SETUP_STOP_COMPLETE   (repeat on)

Supported independently by the THOR manual, which says a repeating schedule
hitting a Start Monitoring event while already monitoring will "stop the current
monitoring session, run any Auto Call Home actions, load the compliance setup and
continue monitoring" -- exactly what _STOP_COMPLETE should mean.  A repeating
start must terminate the in-progress session; a one-shot start need not.  The
firmware name, the manual's behaviour and the single moved byte all agree.

Explicitly NOT claiming the enum ordering: the action names were recovered with
`strings | sort`, so source order is lost.  Do not infer 1 = START_MONITOR just
because it falls between the two known values.  Three of six codes observed.

Also noted: Thor re-pushed the whole 2,090-byte config for a one-byte schedule
change -- a third instance of the schedule<->config coupling.

Capture provenance: the seismo_lab bins did not reach the dev box, but the socat
relay on mint-mac keeps its own timestamped -x log, and the session was
reconstructed from it byte-for-byte (25 frames each way, 0 bad checksums).  That
backup log is worth keeping in the loop -- it has now saved a capture once.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ru8Lg9HkkYvX9VWWo65SmL
2026-09-25 00:35:09 -04:00
..
2026-02-24 21:19:40 +00:00